IceProbe.

[TerraMex]

A friend of mine gave me this link :

http://www.coolworksinc.com/iceprobe.htm

Which i traced back to :

http://www.hardocp.com/
http://www.hardocp.com/image.html?image=MTA0ODgwNzIwMGQ5d1Nuc2RqTTdfMV8xX 2wuanBn

It almost reminds me of that thermaltake activecooling kit, which isnt very good. Having a limited knowledge of pelts, what do you guys think of it?

[bigben2k]

Since it's an air-cooled pelt, it can't be rated for more than 80 Watts, which would handle a heatload of maybe 50 or 60 Watts, but in this configuration, since it doesn't connect to the CPU, but in-line, there will be an added resistance, so I would guess that it might be ok for a CPU that's 30 or 40 Watts (hey, don't quote me on this!).

It's a cutsy use of a round pelt.

It reminds me of the www.overclex.com res, rev 1.0: same principle, using 2 * 80 W pelts, air-cooled. That's a nice weekend-fun project. Actually functional too!

[TerraMex]

http://www.overclex.com/content.php3?id=%2Fwatercooling%2Farticle.php3%3Fi d%3D50%26page%3D1

This one? It does look interesting.

But the main difference that i saw was the use of two 80w rated pelts instead of one. Wouldnt the initial setup, with the IceProbe, eventually heat up in a 24/7 use ?... and maybe to a dangerous level?

If it's rated, and lets be optimistic , 50W , then a full use of the CPU would eventually heat up the water, in a way that the Probe couldnt keep up? If so, adding the pump heat, maybe a NB and graphs wb, would really be disasterous.

And if its subambient temps , mounting a radiador in series with it wouldnt help .

[Skulemate]

It is rated as <50W as far as I can tell from the website. I think it might make a neat addition to a reservoir to help drop temps, but you'd still need a heater-core in there first to remove some heat from the system I imagine... with this alone it would be quickly overwhelmed by most CPUs.

However, if one were to get two or three of their cooling tanks (~60W each) for use in a parallel setup, it might be pretty good in terms of performance. But price, and available space would wreck this plan unfortunately.

[koslov]

You'd get much better efficiency by attaching the aircooled pelts to a WB versus a res. Putting a bunch of 24v pelts running at 12v with good copper heatsinks on a long, rectangular waterblock would probably get some pretty good results. For me though, this would defeat the purpose of WCing: getting rid of those noisy fans!

[Alchemy]

Originally posted by Skulemate
It is rated as <50W as far as I can tell from the website. I think it might make a neat addition to a reservoir to help drop temps, but you'd still need a heater-core in there first to remove some heat from the system I imagine... with this alone it would be quickly overwhelmed by most CPUs.

Much lower. It is rated as *consuming* 50W.

It also states "capable of cooling 10 gallons of water more than 20F under ambient temperature," which means next to nothing. What it implies, though, is that the highest temperature differential is 20F below ambient.

The fact it does this with 10 gallons of water is totally irrelevant; I imagine they meant to give a heat rate but forgot to mention a time.

The charts given help none at all for this application, which seems to be due to the fact they are not marketing to a constant heat load application. The *useful* info would be a heat load vs. temperature differential (ambient vs. probe). This would easily tell you what water temperature you can expect for a given CPU.

Three of these would probably make a badass water cooler, but there are cheaper ways to do it.

Alchemy

[UnloadeD]

Fish tanks do have a heat load on them, pumps, lights, etc. I have no idea what the load would be though. They also have some cooling going on with bubbles and evaporation, all adding to the complexity of the equation. I've looked at these before and did a lot of wondering about them. I can't imagine them in use without a rad. Also the added fan was discouraging. My best idea was to mount one horizontally in a res (they use bulkhead mounting) and replacing the heatsink with a bigger one, and having the fan(s) on my rad blowing on it. I really have no idea if it would do the trick, even if it would there is still the $100 hurdle to overcome.

peace.
unloaded

[Since87]

Originally posted by Alchemy
Much lower. It is rated as *consuming* 50W.


It is a common misconception that TEC's must consume a much greater amount of electrical power than they pump as heat.

There are way to many unknowns here for me to be able to guesstimate the ratio between heat pumping capacity for this thing and power consumption.

However, considering the relatively low dT achieved, it is quite possible that it can pump more than 50 Watts.

(I'm not suggesting this thing might be of any real use for CPU cooling.)

[Alchemy]

Originally posted by Since87
It is a common misconception that TEC's must consume a much greater amount of electrical power than they pump as heat.


Do you have any examples of TECs with Qmax/I^2R better than, say, 0.75? I'd consider anything more than a 25% difference "much lower."

Alchemy

[Since87]

Originally posted by Alchemy
Do you have any examples of TECs with Qmax/I^2R better than, say, 0.75? I'd consider anything more than a 25% difference "much lower."

Alchemy

Examples? How about a Kryotherm simulation?

http://uffish-thought.net/wc-gifs/lowpow.gif

Qc is heat pumped through the cold side of the TEC.
W is power consumed by the TEC.

This simulation shows about a 1:1 ratio between heat pumped and power consumed.